Ewald ALOEBOETOE
ab4bdd41ea
Typing "F" only found file names with a word starting with F (FTS5 prefix match), so the grid did not list e.g. sunflower.png or ComfyUI_00001_.png. Single-word queries now add a filename LIKE '%text%' branch to the global and scoped FTS queries and their counts, ranked after FTS hits, with LIKE wildcards escaped. Multi-word and quoted queries stay pure FTS.
64 lines
2.5 KiB
Python
64 lines
2.5 KiB
Python
"""Single-word searches also match file names that *contain* the text."""
|
|
|
|
from __future__ import annotations
|
|
|
|
from pathlib import Path
|
|
|
|
import pytest
|
|
from mjr_am_backend import deps
|
|
from mjr_am_backend.features.index.searcher import _filename_substring_pattern
|
|
|
|
NAMES = ["ComfyUI_00001_.png", "Flux_portrait.png", "field.png", "sunflower.png", "cat.png", "50%_off.png", "a_b.png", "axb.png"]
|
|
|
|
|
|
def test_pattern_rules():
|
|
assert _filename_substring_pattern("F") == "%F%"
|
|
assert _filename_substring_pattern("red car") is None # multi-word: FTS only
|
|
assert _filename_substring_pattern('"exact"') is None
|
|
assert _filename_substring_pattern("*") is None
|
|
assert _filename_substring_pattern("a_b") == "%a\\_b%" # LIKE wildcards escaped
|
|
assert _filename_substring_pattern("50%") == "%50\\%%"
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_single_letter_lists_every_file_containing_it(tmp_path: Path, monkeypatch):
|
|
root = tmp_path / "out"
|
|
root.mkdir()
|
|
for name in NAMES:
|
|
(root / name).write_bytes(b"\x89PNG\r\n\x1a\n")
|
|
monkeypatch.setattr(deps, "WATCHER_ENABLED", False)
|
|
res = await deps.build_services(str(tmp_path / "db.sqlite"))
|
|
assert res.ok, res.error
|
|
services = res.data
|
|
try:
|
|
index = services["index"]
|
|
scan = await index.scan_directory(
|
|
directory=str(root), recursive=False, incremental=True, source="output", root_id=None
|
|
)
|
|
assert scan.ok, scan.error
|
|
|
|
async def names(query: str, *, scoped: bool) -> set[str]:
|
|
if scoped:
|
|
r = await index.search_scoped(
|
|
query, roots=[str(root)], limit=50, offset=0, filters=None, include_total=True
|
|
)
|
|
else:
|
|
r = await index.search(query, limit=50, offset=0, filters=None, include_total=True)
|
|
assert r.ok, r.error
|
|
found = {a["filename"] for a in r.data["assets"]}
|
|
assert r.data.get("total") in (None, len(found))
|
|
return found
|
|
|
|
for scoped in (True, False):
|
|
assert await names("F", scoped=scoped) == {
|
|
"ComfyUI_00001_.png", "Flux_portrait.png", "field.png", "sunflower.png", "50%_off.png",
|
|
}
|
|
assert await names("flow", scoped=scoped) == {"sunflower.png"}
|
|
# "_" and "%" are literal, not LIKE wildcards.
|
|
assert await names("a_b", scoped=scoped) == {"a_b.png"}
|
|
finally:
|
|
worker = services.get("rating_tags_sync")
|
|
if worker is not None:
|
|
worker.stop()
|
|
await services["db"].aclose()
|